Haveli Restoration Project

Turn a ruin into a restoration story

Take one photograph of an old haveli, fort, or heritage building and produce two things a client actually reacts to: a photoreal restoration proposal that keeps the original structure intact, and a short cinematic video that reveals the transformation from ruin to finished design.

This is the full workflow, written up as a clear step-by-step tutorial. No vague "just use AI" advice. You get the actual prompts, the exact models, and every setting, with reference images at each step so you can see what good looks like before you spend a rupee generating.

What you'll be able to do

• Generate a base image of the existing structure, or start from a real site photo
• Create a proposal that completes only the ruined parts while preserving the heritage fabric
• Explore different design languages (red terracotta jaali, stone and glass, concrete and steel) by changing a few words
• Produce a reveal video that morphs from ruin to finished proposal like a construction time-lapse
• Stitch the clips into one shareable reel

What's inside

• A 10-page illustrated tutorial PDF
• Every prompt, copy-paste ready
• The full model list with exact settings, including the two small mistakes that quietly break video generation
• Reference images for every step

Who it's for

Architects, architectural visualizers, interior and restoration designers, design students, and content creators who want heritage and adaptive-reuse visuals without a full 3D modeling pipeline.

What you need

Access to Claude and Magnific. Both are separate subscriptions. This guide teaches the method and hands you the prompts. It does not include the tool subscriptions themselves.

Buyer Instructions

After purchase, download the tutorial PDF. Open it next to Claude and your Magnific account, then follow the steps in order, starting with the base image. Copy each prompt straight from the guide. The models, settings, and reference images are all in the document. Do one building and one proposal first before attempting the full video reel.